Forecasting device usage using motifs of univariate time-series dataset and combined modeling

Abstract

Systems and methods are provided for compressing a time-series dataset from a monitored device into a compressed dataset representation. Using an unsupervised machine learning model, the system may group a contiguous set of datapoints of the time-series dataset and group, using a distance algorithm, the first cluster to a first motif. Many motifs can be generated to identify different data signatures in the time-series dataset. The plurality of motifs can be used to generate a data definition, motif sequence graph, directed graph, or other combinations of datapoints. These datapoints can be combined through a summation process with other datapoints generated by a second machine learning model. The output of the summation process can be used to forecast device usage of a monitored device in a data center.

What is claimed is: 
     
       1. A system comprising:
 one or more processors; and 
 a machine readable storage medium storing instructions that, when executed by the one or more processors, cause the system to:
 receive original time-series data from a sensor of a monitored device; 
 group the original time-series data into a plurality of clusters; 
 using an unsupervised machine learning model, combine a subset of the plurality of clusters based on data signature similarities to form a plurality of motifs; 
 generate a first data definition using the plurality of motifs, wherein the data definition corresponds with a pre-defined data schema and values for each of the plurality of motifs, wherein the data definition creates a compressed dataset corresponding to the original time-series data; 
 generate a second data definition using a seasonality prediction of the sensor of the monitored device; 
 train a machine learning model to obtain forecasted power consumption data of the monitored device for a future time period with the first data definition and the second data definition, both received from the sensor of the monitored device; and 
 using the forecasted power consumption data obtained by the trained machine learning model, schedule a workload to be executed at the monitored device during a portion of the future time period, wherein the forecasted power consumption data for the portion of the future time period is below a power consumption threshold. 
 
 
     
     
       2. The system of  claim 1 , wherein the first data definition corresponds with a directed graph of the plurality of motifs that comprise weighted edges between data points. 
     
     
       3. The system of  claim 1 , wherein the first data definition and the second data definition are stored as JavaScript Object Notation (JSON) files, and the system further to:
 recover memory space by deleting the original time-series dataset and storing the first data definition and the second data definition in place of the original time-series data. 
 
     
     
       4. The system of  claim 1 , wherein the monitored device is a processor in a server in an IT data center. 
     
     
       5. The system of  claim 1 , wherein the plurality of clusters comprise a local minima and maxima of subsets of the original time-series data. 
     
     
       6. The system of  claim 5 , wherein the system further to:
 using the local minima data point and the local maxima data point from each of the plurality of clusters of the original time-series data, determine a centroid of each of the plurality of clusters. 
 
     
     
       7. The system of  claim 6 , wherein the determination of the centroid of each of the plurality of clusters uses a customized K-Means cluster algorithm that considers a linear fashion of the original time-series data. 
     
     
       8. The system of  claim 1 , wherein the workload that is scheduled to be executed at the monitored device during the portion of the future time period is a reduced workload. 
     
     
       9. The system of  claim 1 , wherein the original time-series data comprises input power periodic data of the monitored device. 
     
     
       10. The system of  claim 1 , wherein the first data definition is generated using a first machine learning model, the second data definition is generated using a second machine learning model, and the first machine learning model and the second machine learning model are different than the trained machine learning model that obtains forecasted power consumption data. 
     
     
       11. The system of  claim 10 , wherein the second machine learning model is one of SARMA, FBProphet, Holt-Winter-ES, or Gated Recurrent Unit Network. 
     
     
       12. A computer-implemented method comprising:
 receiving, by a processor, an original time-series data from a sensor of a monitored device; 
 grouping the original time-series data into a plurality of clusters; 
 using an unsupervised machine learning model, combining a subset of the plurality of clusters based on data signature similarities to form a plurality of motifs; 
 generating a first data definition using the plurality of motifs, wherein the data definition corresponds with a pre-defined data schema and values for each of the plurality of motifs, wherein the data definition creates a compressed dataset corresponding to the original time-series data; 
 generating a second data definition using a seasonality prediction of the sensor of the monitored device; 
 training a machine learning model to obtain forecasted power consumption data of the monitored device for a future time period with the first data definition and the second data definition, both received from the sensor of the monitored device; and 
 using the forecasted power consumption data obtained by the trained machine learning model, scheduling a workload to be executed at the monitored device during a portion of the future time period, wherein the forecasted power consumption data for the portion of the future time period is below a power consumption threshold. 
 
     
     
       13. The computer-implemented method of  claim 12 , wherein the first data definition corresponds with a directed graph of the plurality of motifs that comprise weighted edges between data points. 
     
     
       14. The computer-implemented method of  claim 12 , wherein the first data definition and the second data definition are stored as JavaScript Object Notation (JSON) files, and the method further comprising:
 recovering memory space by deleting the original time-series dataset and storing the first data definition and the second data definition in place of the original time-series data. 
 
     
     
       15. The computer-implemented method of  claim 12 , wherein the monitored device is a processor in a server in an IT data center. 
     
     
       16. The computer-implemented method of  claim 12 , wherein the plurality of clusters comprise a local minima and maxima of subsets of the original time-series data. 
     
     
       17. The computer-implemented method of  claim 16 , wherein the method further comprising:
 using the local minima data point and the local maxima data point from each of the plurality of clusters of the original time-series data, determining a centroid of each of the plurality of clusters. 
 
     
     
       18. The computer-implemented method of  claim 17 , wherein the determination of the centroid of each of the plurality of clusters uses a customized K-Means cluster algorithm that considers a linear fashion of the original time-series data. 
     
     
       19. The computer-implemented method of  claim 12 , wherein the workload that is scheduled to be executed at the monitored device during the portion of the future time period is a reduced workload. 
     
     
       20. The computer-implemented method of  claim 12 , wherein the original time-series data comprises input power periodic data of the monitored device.